Characterizing topography of EDM generated surface by time series and autocorrelation function

Characterizing topography of machined surface is a way of getting insight into the machining phenomena. Here, a statistical procedure is proposed for extracting the topographical features of electric discharge machined (EDM) surface. Computed autocorrelation functions (ACF) of roughness profiles (considering as time series) exhibit possible random and periodic features buried on the machined surface. Through the decomposition of ACF curves and using backward linear prediction method, existing random and periodic patterns are separated. Very small values of a non-dimensional index - PR ratio indicate the presence of significant random variations in the machined surface. Spatial variations of characteristic lengths within a treatment are found as the most contributive part in overall variation and assert stochastic nature of surface development.